It creates new takes, with new backgrounds, textures, lighting, characters and more for video clips up to 10 seconds
Luma AI has added a ‘Modify Video’ function to its Dream Machine application, which alters short video clips, turning them into new takes, with new backgrounds, textures, lighting, characters and more.
It works on video clips up to 10-seconds long, enabling you to “keep what matters and evolve everything else,” says Luma AI.
Modify Video uses advanced performance signals like pose, facial expression, lip sync and scene structure to understand what to preserve and what to reimagine. You can guide the output with visual references, first frame images, or prompts.
According to Luma AI, whether you’re working on previs, pitch decks, or final renders, “you’ll never have to start from scratch again”.
Modify Video preserves motion and action, it provides multiple output variants, you can control outputs with visuals, and you can choose between three presets to define how far the transformation goes.
The three presets are:
Adhere
Closely follows the original video’s look and structure, mainly prioritising retexturing.
Flex
Balances model creativity with input fidelity, for controlled transformations while preserving key elements.
Reimagine
Enables full-scene reinterpretation, including mapping humans to non-humanoid forms like monsters, creatures, or surreal entities.
No comments yet